Trifox, Inc. -- Database Access Experts

Using VORTEXaccelerator

What is the process of authorization when we give a UID/PWD from the Java client to the VORTEXaccelerator on Solaris and VORTEXaccelerator takes a common uid/passwd from the mux.ini so the host connects with this parameter? How do we check the user on the MVS side with the GENESIS user table? What uid/pwd is checked on MVS?: Start up vtxauth.net on Solaris. Then when a client attempts to connect to VORTEXaccelerator on Solaris, VORTEXaccelerator takes the uid/pwd in the connect string and sends a connect request to vtxauth.net which goes to MVS to attempt a GENESIS dbms connection using the uid/pwd. If this succeeds, then the user is allowed to connect to VORTEXaccelerator, otherwise the connection is rejected.
How long is the key buffer for the Hashtable storing SQL statements?: You allocate a maximum amount of memory for the hash when you start the VORTEXaccelerator (hs) and a maximum number of hashed SQL statements (he).
Please tell me how we could set different of users to different MUX group?: TRIM_MUX_NAME differentiates the MUX groups.
What do I have to put in "SHARE ADDR" for the memory segment?: Take MUX8's starting address, E0000000, and add its size (found in decimal at the top of vtxmon's dt display), round it up to the nearest 1M, and make that your new mux's address.
In our environment we only have TRIM_HOME that is VORTEX_HOME? Are they same thing for different version ?: Later releases added VORTEX_* environment variables for customers running VORTEX products without TRIMtools. TRIM_HOME works fine, as discussed in the environment variables chapter of the Trifox Resource Manual.
If I login to the database via the mux there is user checking but no password checking (for the client user, the real user). If I log in directly to the mainframe both user and password get checked. Are there some parameters to start mux with, to enable password checking too ?: Use vtxauth.net. When the user attempts to connect to the VORTEXaccelerator, his userid/passwd is passed to vtxauth.net. vtxauth.net uses the information in net.ini to get the RACF uid/pwd, sends that to MVS, sends the userid/passwd to GENESIS, and if those two succeed, then he is allowed to register with VORTEXaccelerator.
In the Display Monitor screen, what exactly is a dead Client or Slave?: A client or slave process "disappeared" while connected to the VORTEXaccelerator. This could be caused by a dump or a kill sent from another process. The VORTEXdaemon found that the PID in VORTEXacclerator's shared memory area does not exist.
What does 'SLAVE confirm MF' mean?: The slave executed the requested monitor function. This could be a lock, unlock, abdicate, release, etc request
What would cause the Requested and Completed columns to be skewed?: For most entries, timing. For the lock, release entries (any VORTEXmonitor actions that take slave numbers or "*"), the counts may be off if you use "*" instead of actual slave ids.
The M column in the display links screen is vague: The user's manual states that any nonzero value indicates that a monitor function is in progress. Are there other values possible other then 0 and 1, if so what are they?: Yes, there are other values: In addition to 1 (Lock), there are 2 (Release), and 4 (Cancel).
Do any of the initialization parameters have ceilings on them. We found that running on a HP system with different version of Oracle7 the database cursors has a limit of 252. Is there any others?: The only item that is limited is the number of database cursors (252). The other items have no inherent limitations but you may run out of shared memory.
I am using Synergy OpennetServer 6.1.2p on Intel NT 4.0 sp3. Can I use VORTEXaccelerator with this product, if so I am interested in pricing and an evaluation.: Yes you can. Price is $7500/server. Are you sure you need it? On NT we would only recommend it if you plan to have more than 200 connections. Also some databases are better suited than others. Oracle, followed by Informix are the best candidates. Genesis also works well.
The newest fashion of connectstring as I know is something like username/password/adabasc:adabas8. Any other string gives a dbException like "wrong syntax in connectstring (uid/pass/datasource)". That's ok and works really fine, if I don %t use the SUN with accelerator. If I do so, using the same connectstring which works on our NT - machine I'll get the same dbException as above. Is it possible, that vtxhost.mux only sends the connectstring from mux.ini ?: Exactly. Since the VORTEXaccelerator shares slave connections, they must all be connected with the same id/pwd which is read from mux.ini.
Does the same user has to have DBA to update and insert normal tables (non-Trifox ones?): If the userid does not match the owner of the table, then update/insert/delete are not allowed.
How is the "-a" RAC-F checked while using the MUX.: vtxslav.net makes a connection request to vtxnetd on MVS. It uses the information in net.ini to pass the OS (RACF) uid/pwd. At that point, the "-a" RAC-F check is done.
If the number of entries in the hash buffer reaches the max (he) will older sql statements be removed in favor of newer sql statements if the older sql statement have not been used for X period of time? Is there a pseudo-Oracle least recently used algorithm used?: No. Once the hash if filled, no new statements will be cached. The VORTEXaccelerator will continue to operate.
In DT (D - DISPLAY: T - TC cb. (TTC)), what does the total shared memory size, refer to?: The size of the shared memory segment used by this instance of vtxmux.
What happens when you (a) abdicate a slave, (b) free a slave, (c) release/kill a slave, (d) or release/lock a slave in respect to these situations: I. When db has control II. When client has control III. When , on the display links screen, the column W in WFM is set to 1.: Abdicate waits for the slave to be idle before terminating it. Free breaks the link from client to slave unless something is pending on the slave. Kill sends a signal to the slave terminate immediately. Release waits for the slave to be idle before locking it. With regard to your I, II, III, scenarios, kill is the most ruthless; the others wait until the slave is not in use and not in a transaction.
Are the hashed table entries purely SQL statements or are they parsed SQL statements.: Original SQL statements from the application.
Is there any way to find out if the mux group is hung?: Use dm, dl commands within vtxmon and see if they are changing. Check the vtxmux process if it is spinning.